| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <20200721001536.GE19262@shao2-debian>
Date: Tue, 21 Jul 2020 08:15:36 +0800
From: kernel test robot <rong.a.chen@...el.com>
To: Mel Gorman <mgorman@...hsingularity.net>
Cc: Jan Kara <jack@...e.cz>, Amir Goldstein <amir73il@...il.com>,
LKML <linux-kernel@...r.kernel.org>, lkp@...ts.01.org
Subject: [fsnotify] 71d734103e: will-it-scale.per_process_ops 4.2% improvement
Greeting,
FYI, we noticed a 4.2% improvement of will-it-scale.per_process_ops due to commit:
commit: 71d734103edfa2b4c6657578a3082ee0e51d767e ("fsnotify: Rearrange fast path to minimise overhead when there is no watcher")
https://git.kernel.org/cgit/linux/kernel/git/next/linux-next.git master
in testcase: will-it-scale
on test machine: 192 threads Intel(R) Xeon(R) Platinum 9242 CPU @ 2.30GHz with 192G memory
with following parameters:
nr_task: 50%
mode: process
test: eventfd1
cpufreq_governor: performance
ucode: 0x5002f01
test-description: Will It Scale takes a testcase and runs it from 1 through to n parallel copies to see if the testcase will scale. It builds both a process and threads based test in order to see any differences between the two.
test-url: https://github.com/antonblanchard/will-it-scale
In addition to that, the commit also has significant impact on the following tests:
+------------------+---------------------------------------------------------------------------+
| testcase: change | will-it-scale: will-it-scale.per_process_ops 6.4% improvement |
| test machine | 192 threads Intel(R) Xeon(R) Platinum 9242 CPU @ 2.30GHz with 192G memory |
| test parameters | cpufreq_governor=performance |
| | mode=process |
| | nr_task=100% |
| | test=eventfd1 |
| | ucode=0x5002f01 |
+------------------+---------------------------------------------------------------------------+
Details are as below:
-------------------------------------------------------------------------------------------------->
To reproduce:
git clone https://github.com/intel/lkp-tests.git
cd lkp-tests
bin/lkp install job.yaml # job file is attached in this email
bin/lkp run job.yaml
=========================================================================================
compiler/cpufreq_governor/kconfig/mode/nr_task/rootfs/tbox_group/test/testcase/ucode:
gcc-9/performance/x86_64-rhel-8.3/process/50%/debian-10.4-x86_64-20200603.cgz/lkp-csl-2ap3/eventfd1/will-it-scale/0x5002f01
commit:
47aaabdedf ("fanotify: Avoid softlockups when reading many events")
71d734103e ("fsnotify: Rearrange fast path to minimise overhead when there is no watcher")
47aaabdedf366ac5 71d734103edfa2b4c6657578a30
---------------- ---------------------------
%stddev %change %stddev
\ | \
2791175 +4.2% 2909227 will-it-scale.per_process_ops
2.68e+08 +4.2% 2.793e+08 will-it-scale.workload
30001 -1.7% 29506 proc-vmstat.nr_kernel_stack
23.08 ± 10% -16.0% 19.39 ± 17% sched_debug.cfs_rq:/.load_avg.stddev
157.50 ±102% +1470.3% 2473 ± 58% numa-vmstat.node2.nr_inactive_anon
205.75 ± 82% +1149.5% 2570 ± 55% numa-vmstat.node2.nr_shmem
157.50 ±102% +1470.3% 2473 ± 58% numa-vmstat.node2.nr_zone_inactive_anon
8361 ± 6% -14.2% 7175 ± 7% numa-vmstat.node3.nr_kernel_stack
668.50 ± 92% +1380.3% 9896 ± 58% numa-meminfo.node2.Inactive
631.50 ±101% +1466.9% 9895 ± 58% numa-meminfo.node2.Inactive(anon)
773835 ± 5% +12.9% 873656 ± 4% numa-meminfo.node2.MemUsed
826.00 ± 81% +1145.2% 10285 ± 55% numa-meminfo.node2.Shmem
8359 ± 5% -14.1% 7177 ± 7% numa-meminfo.node3.KernelStack
992.25 ± 74% -56.0% 436.50 ± 92% interrupts.33:PCI-MSI.524291-edge.eth0-TxRx-2
15391 ±122% -88.5% 1762 ±127% interrupts.CPU116.LOC:Local_timer_interrupts
992.25 ± 74% -56.0% 436.50 ± 92% interrupts.CPU12.33:PCI-MSI.524291-edge.eth0-TxRx-2
84835 ±146% -96.9% 2626 ±149% interrupts.CPU146.LOC:Local_timer_interrupts
3971 ±129% +237.8% 13415 ± 24% interrupts.CPU148.LOC:Local_timer_interrupts
80671 ±156% -98.9% 916.50 ± 81% interrupts.CPU15.LOC:Local_timer_interrupts
9204 ± 51% +1633.6% 159576 ± 87% interrupts.CPU168.LOC:Local_timer_interrupts
82118 ±153% -96.8% 2591 ±102% interrupts.CPU172.LOC:Local_timer_interrupts
154291 ± 94% -96.8% 4902 ±142% interrupts.CPU186.LOC:Local_timer_interrupts
12237 ± 95% -88.9% 1353 ±132% interrupts.CPU19.LOC:Local_timer_interrupts
467.00 ± 14% -15.5% 394.50 interrupts.CPU190.CAL:Function_call_interrupts
3878 ± 71% +4172.3% 165689 ± 81% interrupts.CPU55.LOC:Local_timer_interrupts
80834 ±156% -98.5% 1228 ± 77% interrupts.CPU59.LOC:Local_timer_interrupts
154405 ± 94% -95.8% 6436 ±133% interrupts.CPU61.LOC:Local_timer_interrupts
14356 ± 47% -33.7% 9516 ± 8% softirqs.CPU116.TIMER
35187 ±109% -70.6% 10336 ± 22% softirqs.CPU146.TIMER
9840 ± 17% +46.3% 14393 ± 21% softirqs.CPU148.TIMER
4744 ± 99% -70.2% 1413 ± 3% softirqs.CPU15.RCU
35797 ±110% -71.7% 10113 ± 2% softirqs.CPU15.TIMER
11678 ± 15% +408.6% 59398 ± 74% softirqs.CPU168.TIMER
3735 ±123% -81.8% 678.25 ± 34% softirqs.CPU172.RCU
35438 ±116% -73.5% 9401 ± 13% softirqs.CPU172.TIMER
6229 ± 84% -87.0% 810.00 ± 44% softirqs.CPU186.RCU
57680 ± 80% -82.9% 9838 ± 22% softirqs.CPU186.TIMER
14204 ± 29% -27.7% 10271 ± 6% softirqs.CPU19.TIMER
36486 ±102% -65.0% 12755 ± 38% softirqs.CPU21.TIMER
5740 ±108% -70.1% 1715 ± 35% softirqs.CPU31.RCU
38176 ±103% -66.4% 12835 ± 31% softirqs.CPU31.TIMER
1353 ± 7% +488.9% 7968 ± 65% softirqs.CPU55.RCU
10688 ± 7% +483.8% 62396 ± 68% softirqs.CPU55.TIMER
36343 ±116% -72.4% 10035 ± 2% softirqs.CPU59.TIMER
7423 ± 77% -80.2% 1470 ± 31% softirqs.CPU61.RCU
58969 ± 78% -80.4% 11528 ± 24% softirqs.CPU61.TIMER
11012 ± 7% +29.5% 14256 ± 12% softirqs.CPU87.TIMER
will-it-scale.per_process_ops
2.96e+06 +----------------------------------------------------------------+
2.94e+06 |-+ O O |
| |
2.92e+06 |-+ O O O O |
2.9e+06 |-+ O O O O O |
2.88e+06 |-+ O |
2.86e+06 |-O O O O O O O O O O O O O |
| O |
2.84e+06 |-+ O |
2.82e+06 |-+ |
2.8e+06 |-+ |
2.78e+06 |.+..+.+.+..+.+. .+.+..+.+ |
| +..+.+.+.. .+..+ |
2.76e+06 |-+ +.+.+..+.+.+ |
2.74e+06 +----------------------------------------------------------------+
[*] bisect-good sample
[O] bisect-bad sample
***************************************************************************************************
lkp-csl-2ap4: 192 threads Intel(R) Xeon(R) Platinum 9242 CPU @ 2.30GHz with 192G memory
=========================================================================================
compiler/cpufreq_governor/kconfig/mode/nr_task/rootfs/tbox_group/test/testcase/ucode:
gcc-9/performance/x86_64-rhel-8.3/process/100%/debian-10.4-x86_64-20200603.cgz/lkp-csl-2ap4/eventfd1/will-it-scale/0x5002f01
commit:
47aaabdedf ("fanotify: Avoid softlockups when reading many events")
71d734103e ("fsnotify: Rearrange fast path to minimise overhead when there is no watcher")
47aaabdedf366ac5 71d734103edfa2b4c6657578a30
---------------- ---------------------------
%stddev %change %stddev
\ | \
1651925 +6.4% 1758026 will-it-scale.per_process_ops
3.172e+08 +6.4% 3.375e+08 will-it-scale.workload
1730 -75.0% 431.75 ±173% meminfo.Mlocked
15.00 +6.7% 16.00 vmstat.cpu.us
815508 ± 4% -8.8% 744090 ± 3% sched_debug.cfs_rq:/.spread0.max
0.12 ± 11% -19.7% 0.09 ± 15% sched_debug.cpu.nr_running.stddev
589.75 ± 4% +18.3% 697.75 ± 6% slabinfo.skbuff_fclone_cache.active_objs
589.75 ± 4% +18.3% 697.75 ± 6% slabinfo.skbuff_fclone_cache.num_objs
109.25 ± 61% -78.9% 23.00 ± 87% numa-numastat.node1.other_node
72.00 ± 96% -95.8% 3.00 ± 23% numa-numastat.node2.other_node
101.00 ± 84% -90.6% 9.50 ±161% numa-numastat.node3.other_node
7551 -1.0% 7478 proc-vmstat.nr_mapped
432.50 -75.1% 107.75 ±173% proc-vmstat.nr_mlock
88431 -1.1% 87446 proc-vmstat.nr_slab_unreclaimable
560.25 ± 28% -29.9% 393.00 interrupts.CPU105.CAL:Function_call_interrupts
3962 ± 67% -53.3% 1850 ± 34% interrupts.CPU145.CAL:Function_call_interrupts
1432 ± 49% +149.3% 3572 ± 21% interrupts.CPU49.CAL:Function_call_interrupts
551.50 ± 35% -28.6% 393.75 interrupts.CPU71.CAL:Function_call_interrupts
12597 ± 6% +12.5% 14177 ± 3% softirqs.CPU130.RCU
12676 ± 2% +16.8% 14811 ± 10% softirqs.CPU131.RCU
13967 +8.0% 15088 ± 4% softirqs.CPU146.RCU
12317 ± 3% +10.4% 13596 ± 5% softirqs.CPU191.RCU
14202 ± 3% +11.1% 15777 ± 7% softirqs.CPU24.RCU
13554 ± 2% +11.5% 15112 ± 4% softirqs.CPU46.RCU
92735 ± 61% +71.3% 158890 ± 36% numa-meminfo.node0.Active
90536 ± 62% +73.9% 157416 ± 38% numa-meminfo.node0.Active(anon)
37320 ±102% +143.2% 90748 ± 51% numa-meminfo.node0.AnonHugePages
89851 ± 62% +75.0% 157225 ± 38% numa-meminfo.node0.AnonPages
7340 ± 53% -64.7% 2592 ±148% numa-meminfo.node0.Inactive
7157 ± 54% -65.7% 2452 ±150% numa-meminfo.node0.Inactive(anon)
8363 ± 23% -29.0% 5936 ± 26% numa-meminfo.node0.Mapped
101341 ± 7% -9.3% 91953 ± 3% numa-meminfo.node0.SUnreclaim
7970 ± 51% -66.0% 2708 ±148% numa-meminfo.node0.Shmem
138005 ± 7% -11.1% 122678 ± 6% numa-meminfo.node0.Slab
22635 ± 62% +73.9% 39357 ± 38% numa-vmstat.node0.nr_active_anon
22463 ± 62% +75.0% 39309 ± 38% numa-vmstat.node0.nr_anon_pages
1789 ± 54% -65.7% 612.75 ±150% numa-vmstat.node0.nr_inactive_anon
2090 ± 23% -29.0% 1483 ± 26% numa-vmstat.node0.nr_mapped
106.50 ± 23% -78.6% 22.75 ±173% numa-vmstat.node0.nr_mlock
1992 ± 51% -66.0% 676.75 ±148% numa-vmstat.node0.nr_shmem
25335 ± 7% -9.3% 22988 ± 3% numa-vmstat.node0.nr_slab_unreclaimable
22635 ± 62% +73.9% 39357 ± 38% numa-vmstat.node0.nr_zone_active_anon
1789 ± 54% -65.7% 612.75 ±150% numa-vmstat.node0.nr_zone_inactive_anon
107.25 ± 22% -78.8% 22.75 ±173% numa-vmstat.node1.nr_mlock
108.00 ± 22% -78.5% 23.25 ±173% numa-vmstat.node3.nr_mlock
Disclaimer:
Results have been estimated based on internal Intel analysis and are provided
for informational purposes only. Any difference in system hardware or software
design or configuration may affect actual performance.
Thanks,
Rong Chen
View attachment "config-5.8.0-rc4-00084-g71d734103edfa" of type "text/plain" (158415 bytes)
View attachment "job-script" of type "text/plain" (7487 bytes)
View attachment "job.yaml" of type "text/plain" (5097 bytes)
View attachment "reproduce" of type "text/plain" (340 bytes)
Powered by blists - more mailing lists